Out of academic curiosity, does anyone know if scalar construction is likely to be more or less expensive than scalar modification? As a matter of fact, it would be really interesting to see performance analyses of many common perl operations.. Maybe they're already out there.. Any pointers, anyone?